2 * Twofish Cipher 8-way parallel algorithm (AVX/x86_64)
4 * Copyright (C) 2012 Johannes Goetzfried
5 * <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
7 * Copyright © 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
26 #include "glue_helper-asm-avx.S"
28 .file "twofish-avx-x86_64-asm_64.S"
34 .byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
38 /* structure of crypto context */
46 /**********************************************************************
48 **********************************************************************/
100 #define lookup_32bit(t0, t1, t2, t3, src, dst, interleave_op, il_reg) \
101 movzbl src ## bl, RID1d; \
102 movzbl src ## bh, RID2d; \
104 movl t0(CTX, RID1, 4), dst ## d; \
105 movl t1(CTX, RID2, 4), RID2d; \
106 movzbl src ## bl, RID1d; \
107 xorl RID2d, dst ## d; \
108 movzbl src ## bh, RID2d; \
109 interleave_op(il_reg); \
110 xorl t2(CTX, RID1, 4), dst ## d; \
111 xorl t3(CTX, RID2, 4), dst ## d;
113 #define dummy(d) /* do nothing */
115 #define shr_next(reg) \
118 #define G(gi1, gi2, x, t0, t1, t2, t3) \
119 lookup_32bit(t0, t1, t2, t3, ##gi1, RGS1, shr_next, ##gi1); \
120 lookup_32bit(t0, t1, t2, t3, ##gi2, RGS3, shr_next, ##gi2); \
122 lookup_32bit(t0, t1, t2, t3, ##gi1, RGS2, dummy, none); \
125 lookup_32bit(t0, t1, t2, t3, ##gi2, RGS1, dummy, none); \
129 #define round_head_2(a, b, x1, y1, x2, y2) \
130 vmovq b ## 1, RGI3; \
131 vpextrq $1, b ## 1, RGI4; \
133 G(RGI1, RGI2, x1, s0, s1, s2, s3); \
134 vmovq a ## 2, RGI1; \
135 vpextrq $1, a ## 2, RGI2; \
137 vpinsrq $1, RGS3, x1, x1; \
139 G(RGI3, RGI4, y1, s1, s2, s3, s0); \
140 vmovq b ## 2, RGI3; \
141 vpextrq $1, b ## 2, RGI4; \
143 vpinsrq $1, RGS3, y1, y1; \
145 G(RGI1, RGI2, x2, s0, s1, s2, s3); \
147 vpinsrq $1, RGS3, x2, x2; \
149 G(RGI3, RGI4, y2, s1, s2, s3, s0); \
151 vpinsrq $1, RGS3, y2, y2;
153 #define encround_tail(a, b, c, d, x, y, prerotate) \
161 vpslld $(32 - 1), c, c; \
165 #define decround_tail(a, b, c, d, x, y, prerotate) \
174 vpslld $(32 - 1), d, d; \
177 #define rotate_1l(x) \
179 vpsrld $(32 - 1), x, x; \
182 #define preload_rgi(c) \
186 #define encrypt_round(n, a, b, c, d, preload, prerotate) \
187 vbroadcastss (k+4*(2*(n)))(CTX), RK1; \
188 vbroadcastss (k+4*(2*(n)+1))(CTX), RK2; \
189 round_head_2(a, b, RX0, RY0, RX1, RY1); \
190 encround_tail(a ## 1, b ## 1, c ## 1, d ## 1, RX0, RY0, prerotate); \
192 encround_tail(a ## 2, b ## 2, c ## 2, d ## 2, RX1, RY1, prerotate);
194 #define decrypt_round(n, a, b, c, d, preload, prerotate) \
195 vbroadcastss (k+4*(2*(n)))(CTX), RK1; \
196 vbroadcastss (k+4*(2*(n)+1))(CTX), RK2; \
197 round_head_2(a, b, RX0, RY0, RX1, RY1); \
198 decround_tail(a ## 1, b ## 1, c ## 1, d ## 1, RX0, RY0, prerotate); \
200 decround_tail(a ## 2, b ## 2, c ## 2, d ## 2, RX1, RY1, prerotate);
202 #define encrypt_cycle(n) \
203 encrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l); \
204 encrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l);
206 #define encrypt_cycle_last(n) \
207 encrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l); \
208 encrypt_round(((2*n) + 1), RC, RD, RA, RB, dummy, dummy);
210 #define decrypt_cycle(n) \
211 decrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l); \
212 decrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l);
214 #define decrypt_cycle_last(n) \
215 decrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l); \
216 decrypt_round((2*n), RA, RB, RC, RD, dummy, dummy);
218 #define transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
219 vpunpckldq x1, x0, t0; \
220 vpunpckhdq x1, x0, t2; \
221 vpunpckldq x3, x2, t1; \
222 vpunpckhdq x3, x2, x3; \
224 vpunpcklqdq t1, t0, x0; \
225 vpunpckhqdq t1, t0, x1; \
226 vpunpcklqdq x3, t2, x2; \
227 vpunpckhqdq x3, t2, x3;
229 #define inpack_blocks(x0, x1, x2, x3, wkey, t0, t1, t2) \
230 vpxor x0, wkey, x0; \
231 vpxor x1, wkey, x1; \
232 vpxor x2, wkey, x2; \
233 vpxor x3, wkey, x3; \
235 transpose_4x4(x0, x1, x2, x3, t0, t1, t2)
237 #define outunpack_blocks(x0, x1, x2, x3, wkey, t0, t1, t2) \
238 transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
240 vpxor x0, wkey, x0; \
241 vpxor x1, wkey, x1; \
242 vpxor x2, wkey, x2; \
246 .type __twofish_enc_blk8,@function;
251 * RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: blocks
253 * RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2: encrypted blocks
262 inpack_blocks(RA1, RB1, RC1, RD1, RK1, RX0, RY0, RK2);
265 inpack_blocks(RA2, RB2, RC2, RD2, RK1, RX0, RY0, RK2);
275 encrypt_cycle_last(7);
277 vmovdqu (w+4*4)(CTX), RK1;
283 outunpack_blocks(RC1, RD1, RA1, RB1, RK1, RX0, RY0, RK2);
284 outunpack_blocks(RC2, RD2, RA2, RB2, RK1, RX0, RY0, RK2);
289 .type __twofish_dec_blk8,@function;
294 * RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2: encrypted blocks
296 * RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: decrypted blocks
299 vmovdqu (w+4*4)(CTX), RK1;
304 inpack_blocks(RC1, RD1, RA1, RB1, RK1, RX0, RY0, RK2);
307 inpack_blocks(RC2, RD2, RA2, RB2, RK1, RX0, RY0, RK2);
317 decrypt_cycle_last(0);
319 vmovdqu (w)(CTX), RK1;
324 outunpack_blocks(RA1, RB1, RC1, RD1, RK1, RX0, RY0, RK2);
325 outunpack_blocks(RA2, RB2, RC2, RD2, RK1, RX0, RY0, RK2);
330 .global twofish_ecb_enc_8way
331 .type twofish_ecb_enc_8way,@function;
333 twofish_ecb_enc_8way:
342 load_8way(%rdx, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
344 call __twofish_enc_blk8;
346 store_8way(%r11, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
351 .global twofish_ecb_dec_8way
352 .type twofish_ecb_dec_8way,@function;
354 twofish_ecb_dec_8way:
363 load_8way(%rdx, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
365 call __twofish_dec_blk8;
367 store_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
372 .global twofish_cbc_dec_8way
373 .type twofish_cbc_dec_8way,@function;
375 twofish_cbc_dec_8way:
387 load_8way(%rdx, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
389 call __twofish_dec_blk8;
391 store_cbc_8way(%r12, %r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
398 .global twofish_ctr_8way
399 .type twofish_ctr_8way,@function;
406 * %rcx: iv (little endian, 128bit)
414 load_ctr_8way(%rcx, .Lbswap128_mask, RA1, RB1, RC1, RD1, RA2, RB2, RC2,
417 call __twofish_enc_blk8;
419 store_ctr_8way(%r12, %r11, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);